1. Field of the Invention
The present invention relates to a method of manufacturing a photo mask, a mask pattern shape evaluation apparatus, a method of judging a photo mask defect corrected portion, a photo mask defect corrected portion judgment apparatus, and a method of manufacturing a semiconductor device.
2. Description of the Related Art
As one cause for a yield drop in manufacturing a large-scaled integrated circuit (LSI), a pattern on a photo mask does not have a desired shape, for use in manufacturing a device by a photolithography technique. There is a so-called defect. Therefore, to manufacture the photo mask, an allowable value is determined beforehand as a specification with respect to a magnitude of the defect. In a step of inspecting the presence of the defect, it is confirmed that there is not any defect having a size which is not less than the value.
In recent years, as miniaturization of a design rule of LSI has advanced, precision required for the photo mask has been rapidly enhanced. For example, a specification value concerning the size of the defect has a level below 0.1 μm. However, the size of the defect determined by the specification value is generally defined assuming a severest condition, when estimating an influence degree in a photolithography step during the transferring onto the wafer, or an influence degree in operating the device.
Therefore, among defects detected in a step of inspecting the presence of the defect, some of them do not raise any problem in actually transferring the pattern onto the wafer. When the pattern on the photo mask including a defect portion is transferred onto the wafer, a shape of the formed pattern falls in an allowable range with respect to the desired shape in a case where the defects do not cause any problem.
Therefore, a method has been developed in which the influence degree is estimated by simulation at a time when the defect detected on the photo mask is transferred to the wafer, and it is judged whether or not the defect has such a size that causes the problem. Accordingly, the necessity of correcting the defect is judged.
Examples of input information during the simulation include an optical image of a defect inspection device for a photo mask, a secondary electron image, an atomic force microscope (AFM) image and the like. When a representation unit of two-dimensional data is considered, only stepwise information is obtained by a finite unit with respect to any image (hereinafter, this representation unit will be referred to as pixel resolution). For example, in the secondary electron image, the pixel resolution comprises a minimum unit determined by magnification of an optical electron system and size of a secondary electron detector.
When estimating the influence degree at the time when the defect on the photo mask is transferred onto the wafer by the simulation, input information preferably has a unit as small as possible. In recent years, by progress of miniaturization, it has been difficult to obtain a pattern intended by a designer on the wafer, when drawing the pattern shape on the photo mask in a simple rectangular figure by photolithography. To solve the problem, the pattern on the photo mask is corrected in such a manner as to finely change, so that the pattern on the wafer has a desired shape in many cases. Therefore, the pattern on the photo mask has many micro stepped portions or micro figures, and a part of the pattern is changed by a unit of 1 nm in some case.
On the other hand, when the influence degree is estimated at the time when the pattern on the photo mask is transferred onto the wafer by the simulation, a certain degree of broadness of the region is required. This is because light passed through a certain point on the photo mask reaches the wafer with a certain range of spread by diffraction phenomenon by a pattern edge. Calculation of the influence from the periphery requires not only the pattern shape of the region whose influence degree is to be estimated but also the pattern shape of the peripheral region. The required range of the peripheral region differs with the shape of the pattern, density, or condition in transferring the pattern onto the wafer, but is considered in a range of 16 μm at maximum on the photo mask.
That is, a combination is supposed to be required in which an image of a 16 μm region be acquired with a pixel resolution of about 1 nm. Considering the present situation of the device for acquiring the image, the number of pixels per side of an image is obtained as follows in a case where the image region acquired at the pixel resolution of 1 nm is 16 μm:16×1000/1=16000.
There has not existed a device capable of acquiring an image having a large number of pixels in the present situation.
From this situation, in a conventional technique, in general, since a region obtained as the image is small, the same pattern is repeatedly generated around the pattern of the range obtained as the image, and a broad region is obtained in a pseudo manner to perform the simulation.
This technique is effective in a region in which patterns regularly continue. However, in a region where there is not regularity, a difference is made between estimated result and actual influence. Alternatively, even in the region where the patterns regularly continue, a difference is made between the pattern in performing the simulation and the actual pattern in an end portion. Therefore, there has been a problem that a difference is made between the estimated result and the actual influence. This difference cannot be ignored in a case where, in steps of manufacturing the photo mask, the influence degree is estimated at the time when the defect on the photo mask is transferred onto the wafer by simulation, and it is judged whether or not the defect needs to be corrected.
FIGS. 18A, 18B, 18C are diagrams of patterns input into simulation in a conventional technique, FIG. 18A is a diagram schematically showing a pattern including a defect on the photo mask, FIG. 18B is a diagram showing a pattern extracted from an image obtained by photographing the vicinity of the defect of FIG. 18A, and FIG. 18C is a diagram showing a pattern input into the simulation, produced by repeatedly arranging the pattern of FIG. 18B.
In FIGS. 18A, 18B, 18C, 1a denotes a transmitting region on the photo mask, 1b denotes a transmitting region in contour extraction data from an image, 2a denotes a shielding region on the photo mask, 2b denotes a shielding region in the contour extraction data from the image, 3a denotes a defect portion on the photo mask, 3b denotes a defect portion in the contour extraction data from the image, 4 denotes a pattern image acquisition region on the photo mask, 5 denotes a required region of simulation input data, and 6 denotes a boundary of pattern repetition.
It is assumed that, as shown in FIG. 18A, the pattern on the photo mask to be simulated is an end portion of a cell pattern of a semiconductor device, the region 4 obtained as the image is smaller than the region 5 required for the simulation, and there is an only pattern in a range in which regularity is held in the obtained image as shown in FIG. 18B. In this case, as shown in FIG. 18C, the pattern to be input into the simulation is prepared in a conventional method for repeatedly producing the pattern data obtained from the image. The pattern is largely different from the pattern on the actual photo mask shown in FIG. 18A, and an error is made in the result obtained by the simulation.
Moreover, similar methods of manufacturing the photo mask are described in Jpn. Pat. Appln. KOKAI Publication Nos. 9-297109, 2000-122265, and 2000-182921.
Furthermore, in the steps of manufacturing the photo mask, the defect on the photo mask detected by a defect inspection apparatus has heretofore been corrected by a defect correction apparatus. The defect portion is usually guaranteed by a microscope AIMS having a light source having a wavelength equal to that of a scanner, and an optical system (NA, σ, illumination condition).
FIG. 19 is a schematic diagram showing a configuration of the AIMS which is an AIMS for a KrF scanner. In this AIMS, first a σ and various illumination aperture 203, and an NA adjustment mechanism 206 are adjusted on exposure conditions for use in the actual scanner. Next, only light having a KrF wavelength (248 nm) is taken out from light emitted from a mercury lamp in a lamp house 201 using an interference filter 202. This light passes through the σ and various illumination aperture 203, and enters a photo mask to be inspected, laid on an XY stage 204.
The light passed through the photo mask to be inspected passes an objective lens 205 and the NA adjustment mechanism 206, and enters a CCD camera 207 for taking in an image. The same light intensity distribution as that transferred onto the wafer is obtained on a light receiving face of the CCD camera 207 for taking in the image, and the light intensity distribution is AD-converted. An image transferred onto the wafer can be obtained by the photo mask to be inspected on the XY stage 204.
From the image transferred onto the wafer, obtained by the AIMS, it is judged whether a defect corrected portion passes/fails on the photo mask to be inspected. In general, it is judged whether the portion passes/fails by judging whether or not the dimension of the corrected portion is within an allowable range with respect to a targeted dimension. It is judged whether the portion passes/fails by judging whether a ratio of the dimension of the corrected portion is not more than a predetermined value with respect to the dimension of a normal portion.
However, in this judgment method, it is necessary to modify the AIMS or introduce a new AIMS simultaneously with the progress of the scanner. For example, when a scanner is released having high NA or new illumination conditions, it is necessary to modify the AIMS accordingly. At present, a mainstream of the scanner is a scanner having a KrF or ArF light source. However, when the scanner of an F2 light source is applied to a product, it is necessary to newly develop and introduce an AIMS having the F2 light source. In the steps of manufacturing the photo mask, a step of guaranteeing the corrected portion is required. It is considered that delay of a mask TAT, and rise of mask cost are also caused.
Moreover, this type of method of correcting the defect is described in Jpn. Pat. Appln. KOKAI Publication No. 2002-14459, and 2002-323749.